Amazon hopes to challenge Nvidia more directly by selling its AI chips
AWS is in active talks to sell its custom AI chips (Trainium/Inferentia) to third-party data centers, with CEO Andy Jassy framing it as a $50B revenue opportunity. This marks Amazon's first serious attempt to monetize its silicon externally and creates a direct competitive dynamic with Nvidia's dominant GPU business.
What happened
AWS is in active talks to sell its custom AI chips (Trainium/Inferentia) to third-party data centers, with CEO Andy Jassy framing it as a $50B revenue opportunity.
The read · Jun 18
The question for AMZN and NVDA is whether AWS can convert internal chip competency into external market share that meaningfully dents Nvidia's 71% gross-margin GPU franchise.
Wrong if
NVDA re-accelerates on Blackwell demand exceeding supply, or AWS chip talks stall with no external customer signed — either outcome collapses the pair thesis.
CoverageSource: TechCrunch · Published here THU, JUN 18 · 2:22 PM ET · the only report in this recordHow this is decided →
The chart · NVDATradingView · third-party feed, not the Wire’s licensed closes
🔒 Click to interact · scroll moves the page
The story1 min read
Amazon Web Services is reportedly in discussions to supply its proprietary AI chips — Trainium and Inferentia — to external data center operators, moving beyond internal use for the first time. CEO Andy Jassy has sized the addressable market at $50B, and given AMZN's $716.9B revenue base growing at 12.4% YoY, even partial capture would be meaningful; however, the chip business has no disclosed revenue contribution yet and faces a long ramp from zero external traction.
Nvidia's moat remains formidable: NVDA posted 65.5% revenue growth to $215.9B with 71.1% gross margins — figures that reflect deep software lock-in via CUDA, not just hardware superiority. The key watch items are whether AWS can sign any flagship external customer to validate the $50B framing, and whether Nvidia's margins show any compression in upcoming prints as hyperscaler in-house silicon gains credibility.
The read, in fullWhy it matters
The pair trade (long AMZN / short NVDA) is grounded in the asymmetry between AMZN's early-stage external chip ambition — a real but unproven $50B framing — and NVDA's 71.1% gross margins, which are priced for continued dominance and are the most exposed if hyperscaler alternatives gain traction. However, there is no confirmed external AWS chip customer yet, limiting the near-term catalyst for AMZN and reducing conviction on NVDA downside.
The arc● Open · no developments yet
· Published
The read above, as written. kept as written · closes shown from JUN 18 on
Next · what would move this
A dated catalyst on AUG 27 · 4-8 weeks, into next NVDA earnings print. Follow to be told when one lands.
Since this story · named here, equal weight · 1D EOD+7.6%
JUN 18 · first close after publicationSEP 4
Price context does not establish that the story caused the move.
▲ The case it holds
AMZN's $50B framing, combined with cost-sensitive data center operators seeking Nvidia alternatives, could catalyze a re-rating of AWS's silicon segment at a time when AMZN net margins (10.8%) have meaningful upside leverage from high-margin chip licensing.
▼ The case it breaks
Nvidia's 65.5% revenue growth and CUDA software lock-in represent a durable moat that AWS chip efforts have so far not cracked — Amazon has been building custom silicon for years with zero disclosed external revenue to show for it, making the $50B opportunity speculative at this stage.
Kept as written · your side, if you take one, is graded privately against licensed closes after 10 trading days · nothing here is advice · How the Wire is made →
Shares a name with this story — discovery, not a connection.
The memory
This page is kept as it was written on Jun 18. Later coverage joins it only when the company and catalyst evidence match, and what the stock did is shown from licensed end-of-day closes — never re-graded, never backdated. The judgment is yours.